Quote:
Originally Posted by Lil' Lavery
My point was not saying that it would work in FRC, but to merely reference the results of non-random schedules in FRC.
|
Bill's method does not produce non-random.schedules. Bill's method randomly selects teams to populate (tri) matches in which the 4 randomly selected teams get to enjoy competing in all permutations of 4 teams rather than being stuck with the results of only one of those permutations. To say it another way,
random selection of the teams in matches is combined with the tri_match notion that improves the tournament's ability to identify valid alliance captains.
Quote:
Originally Posted by Lil' Lavery
Once again, the FIRST community reacted with disgust to the non-random features, and the inability to play against a wide variety of different opponents.... each team would only get to interact with either 3 or 6 other teams during the entire qualification rounds with your system.
|
See my other message below. Are my assumptions reasonable? Is my arithmetic correct? Do the results lead you to rethink this objection?
Quote:
Originally Posted by Lil' Lavery
My general point about defense is that the defender's "ranking" within the group would vary depending on their opponents.
|
I agree; and I agree that there is a further complication because a defender's effect on any one match strongly depends on the scoring method of the robot they are harassing. However, I am not convinced that this will change the net result of the Bill's original simulation.
To assert that, one either needs to do the math or run an adequate simulation. No one has posted the results of doing either.
At the worst, I think that accurately incorporating defense into the simulation will reduce but not eliminate its advantages. At best I think the advantages will continue to be substantial. These are my opinions, and like all of our opinions on the subject, they have limited usefulness without some hard facts to back them up.
Quote:
Originally Posted by Lil' Lavery
Everyone has a problem with "false" captains ranking in the Top 8, but I don't believe that this is the solution. Even if it creates more accurate rankings ... you sacrifice too much in interaction with other teams to make it worth it.
|
See my other message below. Are my assumptions reasonable? Is my arithmetic correct? Do the results lead you to rethink this objection?
Quote:
Originally Posted by Lil' Lavery
The most reliable way, under any format (and your data shows this too) to increase the accuracy of the rankings is to play more matches.
|
If I recall Bill's results correctly, dramatically increasing the number of matches in a tournament run using last season's methods had far less effect on the rankings' accuracy than the effect of switching to Bill's suggested method.
Quote:
Originally Posted by Lil' Lavery
I think our efforts would be better focused on how to make competitions more streamlined and quicker running in order to have more qualification matches.
|
I think you are up against a diminishing-returns situation here. See my comment immediately above and see my other message below. Are my assumptions reasonable? Is my arithmetic correct? Do the results lead you to rethink this objection?
Quote:
Originally Posted by Lil' Lavery
... given your location in Virginia and relationship with Blake...
|
Bill and I have spoken in person for about 15 minutes once and have exchanged a few emails. You made a good guess, but are mistaken. He and I aren't close friends. I simply like the results of his math and the carefully-described, easily-repeated, easily-tested method he used to develop them. Future scientists, engineers, and business managers/owners should learn how to apply the same methods and rigor.
Quote:
Originally Posted by Lil' Lavery
I assume you attended the Farmville competition, given your location in Virginia ... where there were only three qualification matches for each team. Most other competitions got more than this, causing more accurate results.
|
I don't remember the exact number but I am pretty sure that teams at the Virginia FVC Championship (in Farmville) were able to participate in more than 3 qual matches. I think the number was more like 5 or 6.
Regardless, Bill's team went to another championship event and to the FVC World Championship, and I don't think his suggestion is prejudiced too much by his experiences at any one of those events.
Blake
(still enjoying the debate) Ross